Instructions to use tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO")
model = AutoModelForCausalLM.from_pretrained("tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO

SGLang

How to use tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO with Docker Model Runner:
```
docker model run hf.co/tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO
```

MedQA_L3_350steps_1e7rate_03beta_CSFTDPO

This model is a fine-tuned version of tsavage68/MedQA_L3_1000steps_1e6rate_SFT on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 0.6516
Rewards/chosen: 0.2738
Rewards/rejected: 0.1790
Rewards/accuracies: 0.7099
Rewards/margins: 0.0948
Logps/rejected: -33.2582
Logps/chosen: -30.4158
Logits/rejected: -0.7313
Logits/chosen: -0.7305

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 1e-07
train_batch_size: 2
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_steps: 100
training_steps: 350

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6925	0.0489	50	0.6930	-0.0016	-0.0023	0.5011	0.0007	-33.8624	-31.3338	-0.7320	-0.7314
0.6841	0.0977	100	0.6807	0.2459	0.2195	0.6549	0.0264	-33.1233	-30.5088	-0.7330	-0.7323
0.6524	0.1466	150	0.6658	0.3522	0.2898	0.6703	0.0624	-32.8887	-30.1544	-0.7315	-0.7308
0.631	0.1954	200	0.6545	0.1829	0.0948	0.6923	0.0881	-33.5389	-30.7188	-0.7310	-0.7303
0.6675	0.2443	250	0.6520	0.2481	0.1544	0.7121	0.0938	-33.3403	-30.5014	-0.7309	-0.7301
0.6479	0.2931	300	0.6509	0.2738	0.1773	0.7099	0.0966	-33.2640	-30.4157	-0.7310	-0.7303
0.6583	0.3420	350	0.6516	0.2738	0.1790	0.7099	0.0948	-33.2582	-30.4158	-0.7313	-0.7305

Framework versions

Transformers 4.41.1
Pytorch 2.0.0+cu117
Datasets 2.19.1
Tokenizers 0.19.1

Downloads last month: 6

Safetensors

Model size

8B params

Tensor type

F16

Model tree for tsavage68/MedQA_L3_350steps_1e7rate_03beta_CSFTDPO

Base model

meta-llama/Meta-Llama-3-8B-Instruct

Finetuned

tsavage68/MedQA_L3_1000steps_1e6rate_SFT

Finetuned

(20)

this model